Should I set Reaction.Type=On?

I’d like to say thanks to every one spend his/her time on reading my thread.

In part of my modeling process, I should model a zero-mass segment that connects a spring to two wires. I have controlled movement of this segment using a AnyKinEqInterPolDriver. I set Reaction.Type=Off, because I believe movement of this segment governs by spring and ropes. Kinematics ran excellent.
But Inverse Dynamic Analysis fail because of KKT singularity. When I set reaction forces to On, Inverse dynamic runs completely. But in my opinion, It should be Off, am I wrong?
I attached a picture which shows the segment, spring(blue line) and wires(black lines).

1 Like


It is a tricky question, in principle you are right, if you drive the position of the massless segment wrt to ground you are right that the reaction on that driver should be off, but then you might be missing other constraint forces in the model which can then balance the segment. The force in the string and the two spring may not balance the segment.

Please note that even though there should be no force out of the plane (i assume the model to be 2D) you will still need reactions in that direction to be covered, similar for rotational directions, and sideways to the string.

If you look at the force balance in the direction of the string you can switch off the reaction in that direction provided that the string and springs balance out each other, and depending on how you defined the model this needs not to be the case?

If you are ultimatively looking for a model where you pull the string with a force and then you obtain a motion of the segment which are in balance with the springs then you should consider looking at the force dependent solver, it can do exactly that but it takes more time to compute. Please see the example

Best regards


Thanks for your reply. You are right, I can use reaction force only in direction of spring and rope, I mean just one DOF in 6 DOF. Results are better in this situation, but I can’t get the point.
I know the motion of this segment completely (imposed by AnyKinEqInterPolDriver). Assume we denote the direction of spring and ropes and therefore the line in which segment moves by ‘x’. You may right, because there is no force component (from rope or string) along y and z axis and so software can not solve the problem. But in the real world we can see this phenomenon. can’t?

I would be thankful, if you could provide more information on this matter.

Thanks again for explanations.


Hi Majid,

In real life the reactions y and z are zero, they will also be this in the model but they will need to active reactions. The model may have numeric problem if you do not allow these two reactions.

Concerning the z direction it should be off but for that to solve it requires that the force in the line and the spring are in equilibrium, and this may not be the case if the force is applied by a preset value in the script.

If you added an AnyReac force to the line it would solve because then the system coulkd decide on its value needed to balance the two springs.

Alternative take a look at the FDK example it will find the position of the segment that balance out the loads off springs and applied load.

Best regards